Comment by leobg

Interesting. WhisperMemos user here.

I used audio journaling before Whisper came out, and I did have a pipeline that would run a transcription through Google Cloud and save the transcript to Evernote. I didn’t actually review the transcripts most of the time, but the very fact of developing my thoughts—without being constrained by typing speed—was very helpful.

What I also liked about this system was that it gave me independence of place: I didn’t have to sit down at a computer. Instead, I could be thinking aloud while driving a car, or while taking a walk in the countryside. Usually, after finishing such a conversation with myself, I would automatically feel much more clarity about the upcoming day, or about whatever issue that had been on my mind.

If I were to add AI to this process, I would perhaps only have the AI extract some bullet point summaries for the topics that were covered, and anything that could be considered a potential to-do item—so that as output you would get these high-level summaries, along with the raw transcript.

If you wanted to, you could also color-code for each summary item the parts of the raw transcript in which these are covered. So, if you ever do look at the summary-slash-transcript, you can always quickly look up what your thoughts were on the subject—though I would guess this would not happen often.

And the most value from that would probably be that your future self, say 10 or 20 or 50 years from now, will be able to dive into what went through your head at this point in time. For the immediate present, the most value probably just comes from taking the mess that is in your head of unfinished thoughts and serializing it into coherent speech—until you feel everything that’s on your mind has been said.